Uniaxial compressive strength (UCS) of rocks is the most commonly used parameter in geo-engineering application. However, this parameter is hard for measurement due to a time consuming and requires expensive equipment. Therefore, obtaining this value indirectly using non-destructive testing methods has been a frequently preferred method for a long time. In order to obtain multiple regression models, input parameters need many assumptions. Thus, the estimation of the mechanical properties of rocks using by machine learning methods has been investigated. In this study, UCS values of rocks were estimated by reformulating with artificial intelligence-based age-layered population structure genetic programming (ALPS-GP) which is one of machine learning methods. Artificial neural network (ANN) and ALPS-GP models were performed to predict UCS from porosity, Schmidt hammer hardness and ultrasonic wave velocity test methods. For this purpose, the mentioned three tests (porosity, Schmidt hammer hardness and P-wave velocity) were carried out on ten different stones from Turkey. ANN was performed to evaluate this new technique. Reliability of UCS values determined by models was checked with mean absolute error (MAE), coefficient of determination (R²), root mean square error (RMSE) and variance account for (VAF) values. These values were calculated as 1.64, 0.98, 2.11 and 99.81 for ANN, and 2.11, 0.98, 2.50 and 97.86 for ALPS-GP, respectively. It was observed that both methods used were quite successful in UCS estimation. The most important advantage of the ALPS-GP model is providing an equation for UCS estimation. In the light of the obtained findings, it has been revealed that this equation derived from ALPS-GP can be used in UCS estimation processes of similar rock types (limestone, dolomite and onyx).

岩石的单轴抗压强度 (UCS) 是地球工程应用中最常用的参数。然而，该参数由于耗时且需要昂贵的设备而难以测量。因此，长期以来，使用无损检测方法间接获得该值一直是首选方法。为了获得多元回归模型，输入参数需要很多假设。因此，已经研究了使用机器学习方法估计岩石的力学特性。在这项研究中，岩石的 UCS 值是通过重新制定基于人工智能的年龄分层种群结构遗传编程（ALPS-GP）来估计的，这是一种机器学习方法。使用人工神经网络 (ANN) 和 ALPS-GP 模型从孔隙率、施密特锤硬度和超声波速度测试方法预测 UCS。为此，对来自土耳其的十种不同石头进行了上述三项测试（孔隙率、施密特锤硬度和 P 波速度）。执行人工神经网络以评估这种新技术。模型确定的 UCS 值的可靠性通过平均绝对误差 (MAE)、确定系数 (R² )、均方根误差 (RMSE) 和方差解释 (VAF) 值。这些值分别计算为 ANN 的 1.64、0.98、2.11 和 99.81，以及 ALPS-GP 的 2.11、0.98、2.50 和 97.86。据观察，使用的两种方法在 UCS 估计中都非常成功。ALPS-GP 模型最重要的优点是提供了一个用于 UCS 估计的方程。根据所获得的研究结果，表明该由 ALPS-GP 导出的方程可用于类似岩石类型（石灰石、白云石和缟玛瑙）的 UCS 估计过程。